Formalizes regret minimization with free exploration, introduces (α,β)-probably saving policies and UFE-KLUCB-H algorithm, and proves instance-dependent regret savings with upper and lower bounds in the logarithmic budget regime.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
On the Benefits of Free Exploration for Regret Minimization in Multi-Armed Bandits
Formalizes regret minimization with free exploration, introduces (α,β)-probably saving policies and UFE-KLUCB-H algorithm, and proves instance-dependent regret savings with upper and lower bounds in the logarithmic budget regime.